Model Test of a 1:8 Scale Floating Wind Turbine Offshore in the Gulf of Maine1

Abstract

A new floating wind turbine platform design called VolturnUS developed by the University of Maine uses innovations in materials, construction, and deployment technologies such as a concrete semisubmersible hull and a composite tower to reduce the costs of offshore wind. These novel characteristics require research and development prior to fullscale construction. This paper presents a unique offshore model testing effort aimed at derisking fullscale commercial projects by providing scaled global motion data, allowing for testing of materials representative of the fullscale system, and demonstrating fullscale construction and deployment methods. A 1:8scale model of a 6 MW semisubmersible floating wind turbine was deployed offshore Castine, ME, in June 2013. The model includes a fully operational commercial 20 kW wind turbine and was the first gridconnected offshore wind turbine in the U.S. The testing effort includes careful selection of the offshore test site, the commercial wind turbine that produces the correct aerodynamic thrust given the wind conditions at the test site, scaling methods, model design, and construction. A suitable test site was identified that produced scaled design load cases (DLCs) prescribed by the American Bureau of Shipping (ABS) Guide for Building and Classing Floating Offshore Wind Turbines. A turbine with a small rotor diameter was selected because it produces the correct thrust load given the wind conditions at the test site. Some representative data from the test are provided in this paper. Model test data are compared directly to fullscale design predictions made using coupled aeroelastic/hydrodynamic software. Scaled VolturnUS performance data during DLCs show excellent agreement with fullscale predictive models. Model test data are also compared directly without scaling against a numerical representation of the 1:8scale physical model for the purposes of numerical code validation. The numerical model results compare favorably with data collected from the physical model.